Improved Adhesion of Dense Silica Coatings on Polymers byAtmospheric Plasma PretreatmentLinying Cui,†,‡ Alpana N. Ranade,§ Marvi A. Matos,§ Geraud Dubois,*,†,⊥ and Reinhold H. Dauskardt*,†
†Department of Materials Science and Engineering and ‡Department of Applied Physics, Stanford University, Stanford, Califormia94305, United States§Chemical Technology Division, Boeing Research and Technology, Seal Beach, California 90740, United States⊥IBM Almaden Research Center, 650 Harry Road, K-17/E-1, San Jose, California 95120, United States
ABSTRACT: Oxygen atmospheric plasma was used to pretreat polycarbonate (PC) and stretched poly(methyl methacrylate)(PMMA) surfaces in order to enhance the adhesion of the dense silica coatings deposited by atmospheric plasma on the polymersubstrates. The treatment time and chemical structure of the polymers were found to be important factors. For PC, a shorttreatment increased the adhesion energy, while longer treatment times decreased the adhesion. In contrast, plasma pretreatmentmonotonically decreased the adhesion of PMMA, and pristine PMMA exhibited much higher adhesion than the PC counterpart.We found that adhesion enhancement was achieved through improved chemical bonding, chain interdiffusion, and mechanicalinterlocking at the coating/substrate interface, after a short atmospheric plasma treatment. Decreased adhesion resulted fromoveroxidation and low-molecular-weight weak layer formation on the polymer surface by prolonged atmospheric plasmatreatment. The dramatic differences in the behavior of PC and PMMA in relation to the plasma treatment time were due to theirdissimilar resistance to atmospheric plasma exposure.
Polymers are often coated by protective or functional coatings toextend the performance and lifetime of the final product. One ofthe most common coatings on polymers is silica, for the purposeof hard protective coatings,1 permeation barriers to gas diffusion(when dense enough),2 wettability layers (when covered bysilanol groups),3 optical layers in the fabrication of photovoltaicsolar cells,4 and corrosion-resistant layers in precision engineer-ing (aeronautical and automotive).1,5 However, most polymersare hydrophobic because of their major elemental chemicalcomposition of hydrogen and carbon and have poor adhesionwith hydrophilic coatings, including various oxides andnitrides.6,7
To improve the adhesion of hydrophilic coatings on polymers,several surface treatments of polymers have been studied.Mechanical roughening6 and wet chemical treatment8 are amongthe traditional methods. Plasma treatments by corona and low-pressure glow discharges have also been intensively studied andwidely accepted in industry.9,10 In recent years, an atmospheric-
pressure plasma jet was reported for modifying polymersurfaces.11−14 It has been shown that atmospheric plasma canincrease the polymer surface energy by forming alcohol, ketone,aldehyde, carbonate, ester, and ether groups through oxidationreactions.11,12 The surface morphology is also modified aftertreatment.11,13 Because of the chemical and morphologicalmodifications on the surface, the adhesion between the treatedpolymer and epoxy was reported to be significantlyimproved.12,13 However, to our knowledge, there has not beenany report about the important factors and different possiblechanges in the adhesion between an atmospheric-plasma-pretreated polymer and hydrophilic atmospheric plasma coat-ings.The atmospheric plasma process has the advantage of not
requiring vacuum equipment, so the initial capital investment is
Received: May 20, 2013Accepted: August 13, 2013Published: August 13, 2013
largely decreased, and when integrated with other tools, it allowsfor the treatment of large and/or complex geometry substrates.15
A better understanding of this emerging technique for improvingthe surface properties of polymers requires knowledge of bothatmospheric plasma and adhesion mechanisms.The principal adhesion mechanisms to polymers include (1)
chemical interactions (covalent, ionic, etc.), (2) mechanicalinterlocking (surface topography), (3) interdiffusion of chains,and (4) other weak interatomic forces (hydrogen bonds, van derWaals forces, etc.). All of these mechanisms can be modified byatmospheric plasma treatment through the effects of cleaning,etching, chemical modification, and cross-linking of the surface.15
Surface cleaning removes low-molecular-mass contaminants thatinterfere with bonding. Etching cuts the polymer chains to createmore chain ends, which improve interdiffusion between thepolymer and coating species at the molecular scale. At a largerlength scale, etching increases the surface area and promotesmechanical interlocking. Chemical modification is also importantbecause the newly formed polar functional groups can improvecovalent-bonding and intermolecular interactions at the inter-face. Finally, cross-linking the surface polymer layer by ultraviolet(UV) plasma radiation can cohesively strengthen the polymerlayer and potentially enhance its adhesion property.In this study, we used oxygen atmospheric plasma to pretreat
the polycarbonate (PC) and stretched poly(methyl methacry-late) (PMMA) substrates and studied their adhesion variationwith hydrophilic silica coatings deposited by atmospheric plasma.We observed a factor of 4 increase in adhesion to the PCsubstrate with a 30 s atmospheric plasma treatment, whileprolonged treatment decreased the adhesion. In contrast, plasmapretreatment monotonically decreased the adhesion of PMMA,and pristine PMMA exhibited much higher adhesion than the PCcounterpart.To explain the different behavior of the adhesion on PC and
PMMA, surface studies after different amounts of atmosphericplasma exposure were performed by X-ray photoelectronspectroscopy (XPS) and atomic force microscopy (AFM).Two competing effects of atmospheric plasma treatment wereobserved: (1) After short atmospheric plasma exposure, thepolymer surface can be modified with more polar groups andchain ends and higher surface roughness, all of which enhance apolymer’s adhesion to hydrophilic coatings. (2) Plasma exposurecan lead to the formation of a low-molecular-weight layer(LMWL) on the surface, especially with longer treatment. TheLMWL hindered effective interaction between the coating andthe bulk substrate, resulting in adhesion decrease. Interestingly,the surface modification kinetics was strongly correlated with thechemical structure of the polymer, as described in the paper.
■ EXPERIMENTAL METHODSCoating Deposition. An atmospheric pressure plasma system
[Surfx Technologies LLC, Redondo Beach, CA] integrated with a high-temperature precursor delivery system was employed to pretreat theplastic substrates and deposit the coating.16 The area of the plasmashowerhead was 5.1 cm2. A total of 99.995% purity quality helium andoxygen [Praxair Inc., Santa Clara, CA] were mixed and fed into thecapacitive discharge plasma. The plasma was driven by 13.56 MHzradio-frequency (RF) power. Reactive species were generated in theafterglow region of the plasma, including ground-state oxygen atoms(3P), metastable molecular O2 (1Δg and 1∑g+), and ozone.17 UVphotons were also generated in the vicinity of the plasma source.6,18 Itwas reported previously that He/O2 atmospheric plasma can produce avariety of UV photons, such as the O I line at 130 nm and other linesfrom 200 to 400 nm.6,18,19 These photons are sufficiently energetic to
break organic bonds and get absorbed in a polymer surface typically afew tens of nanometers deep.6
A carbon-bridged precursor, 1,2-bis(triethoxysilyl)ethane [Gelest,Inc., Morrisville, PA], was used to deposit silica coatings. The precursorwas vaporized at 120 °C with a vapor pressure of 1.6 Torr.16 Theprecursor bubbler had helium gas flowing through at 0.1 L/min flow rateand 1 atm of pressure. The outcoming gas was a mixture of helium andsaturated precursor vapor. The plasma parameters used were 30 L/minhelium, 0.5 L/min oxygen, and 60WRF power. A detailed description ofthe deposition conditions was given in our previous paper.16 Coatingswere deposited on silicon (100) wafers, PC [Makrolon Ltd., San Diego,CA], and military-grade stretched PMMA sheets meeting all require-ments of MIL-PRF-25690. The substrate was wiped with ethanol beforedeposition to remove any surface contamination and dust and then driedin air for 24 h. The substrate was placed 5 mm below the plasma sourceand exposed to the plasma afterglow. Deposition of a uniform coatingwith controlled thickness was implemented through the use of an X−Y−Z stage that moved the plasma source over the substrate in a planarfashion, forming a rectangular array. The speed of the plasma source was50 mm/s. The spacing between neighboring lines in the array was 0.3mm.
Atmospheric Plasma Treatment of the Stretched PMMA andPC Substrates. The PC and PMMA substrates were pretreated byoxygen atmospheric plasma for different amounts of time in order toactivate the surface before deposition. The same plasma conditions asthose for deposition were used: 30 L/min helium, 0.5 L/min oxygen,and 60 W RF power. The substrate was placed 5 mm below the plasmasource, exposed to the plasma afterglow. An X−Y−Z stage moved theplasma source over the substrate in the same fashion as that described inthe deposition section above. The PC substrates were plasma-treated for0, 15, 30, 140, and 270 s before coating deposition and for 0, 30, 60, and270 s for surface property characterization (XPS and AFM analyses).The PMMA substrates were treated for 0, 5, 17, 34, 170, and 510 s beforecoating deposition and for 0, 2, 5, 10, 60, 180, and 900 s forcharacterization (XPS and AFM analyses).
Characterization Methods. The coating thicknesses werecharacterized by ellipsometry [Woollam M2000; J. A. Woollam Inc.,Lincoln, NE]. Incident light (45° polarization) at the Brewster angle ofthe substrate was used to maximize reflection. Polarization of thereflected light versus wavelength was first taken for the coating on thesilicon substrate. Software was used to fit the refractive index andabsorbance of the coating by regressive analysis given the siliconsubstrate properties. Then a spectrum was taken for the bare polymersubstrate (wavelength 250−1000 nm). Finally, a spectrum was taken forthe coating on the polymer substrate. The coating thickness was fittedbased on the measured refractive index and absorbance of the coatingand substrate spectrum.
Chemical bonding in the coating was characterized using IRspectroscopy. The spectrum was recorded as power dispersions inKBr (reflectance mode) [Nexus 670 FT-IR; Thermo Fisher ScientificInc., Waltham, MA]. Mid-IR (wavelength 400−4000 cm−1) was probedat a resolution of 4 cm−1. Coatings on silicon substrates werecharacterized in transmission mode at the Brewster angle of the siliconsubstrate.
The surface morphology of the coatings and the atmospheric-plasma-pretreated PMMA and PC were characterized by AFM [a Park Systemsmodel XE-70 scanning probe microscope, Park Systems Inc., SantaClara, CA]. Noncontact mode was used, with a scan area of 1.267 μm ×1.267 and 1.7 μm Z range. The root-mean-square surface roughness wasobtained using the XEI software equipped with the AFM machine.
The chemical state of the carbon species on the polymer surface wasanalyzed by XPS [Physical Electronics Inc., Chanhassen, MN] within 10min after atmospheric plasma treatment. An Al Kα (1486 eV) X-raysource was used (spot size ∼1 mm, pass energy 23.5 eV, and scan range20 eV). All of the spectra were referenced to the C 1s peak of thealiphatic C−H/C−C at 285.0 eV. The elemental composition of thecoatings and the fracture surfaces after the adhesion/cohesion test wasalso characterized by XPS (pass energy 117.4 eV and scan range 0−1000eV). Surface contamination was removed by argon-ion beam sputtering
ACS Applied Materials & Interfaces Research Article
before measurement of the coating composition (sputter rate 9 nm/minfor typical PECVD silica coatings and sputter time 5 min).The adhesion energy of the coating on the plastic substrate was
quantified using the asymmetric double-cantilever beam (ADCB)test.1,16,20−22 The specimens were prepared by gluing a blank thinnersubstrate (beam) onto a coated thicker substrate (beam). The in-planedimensions of the specimen were 9 mm × 70 mm. Because of theavailability of the thickness of the plastic sheet, the PC and PMMAspecimens had different thicknesses; the PC specimen had a 4.5-mm-thick blank beam and a 5.9-mm-thick coated beam, and the PMMAspecimen had a 3-mm-thick blank beam and a 6-mm-thick coated beam.The fracture tests were conducted on a micromechanical adhesion testsystem [DTS Delaminator Test System, DTS Co., Menlo Park, CA] indisplacement control mode. The specimens were loaded (displacementrate 5 μm/s) in tension to produce controlled crack growth, followed byunloading. The load was measured simultaneously, and the adhesionenergy Gc (J/m
2) was calculated from the critical value of the strainenergy release rate using1,20−22
=′
+ + +⎡⎣⎢⎢⎛⎝⎜
⎞⎠⎟
⎛⎝⎜
⎞⎠⎟
⎤⎦⎥⎥G
PE B
ah h
ah h
60.64
10.64
1c
c2
21
2
1 2
2
2 (1)
where Pc is the load when the load−displacement curve deviated fromlinearity until the initial crack extension, E′ the plane-strain Young’smodulus of the substrate, B the substrate width, a the crack length, andh1 and h2 the substrate thicknesses. Application of the technique to thinhard coatings on softer substrates of type similar to that in the presentstudy has been previously reported.1,16
■ RESULTS AND DISCUSSIONUniform dense silica coatings of ∼600 nm thickness weredeposited on the PC and PMMA substrates by atmosphericplasma. The adhesion of the coatings was remarkably affected bythe atmospheric plasma pretreatment time and the chemicalstructure of the substrate. XPS and AFM analyses correlated thesurface chemical and morphological changes with the adhesionvariations, revealing two competing effects of atmosphericplasma pretreatment on the polymer’s adhesion.Coating Properties. The deposition rate of the coating was
116± 20 nm/min on PC and 106± 22 nm/min on PMMA. Thesurface roughness of the coatings was below 1 nm on allsubstrates. Characterization of the coatings’ atomic and chemicalproperties as well as the density and Young’s modulus waspreviously reported.16 The silica coating was composed of 5 atom% of carbon, 32 atom % of silicon, and 63 atom % of oxygen byXPS measurement. Hydrogen was not included in the atomiccomposition calculation because of limitations of the technique.The Si−O−Si network structure was formed in the coating, asrevealed by its IR spectrum.16 The carbon residues were in theform of Si−O−C andCO, suggesting that the carbon bridge inthe precursor molecule was oxidized under the currentdeposition conditions. The Si−OH stretching mode was alsoobserved in the IR spectrum, as a common feature for as-deposited low-temperature PECVD silica coatings.23 Thecoating had a density of 1.833 g/cm3 and a Young’s modulusof 22.7 ± 2.3 GPa, much higher than those of the commercialpolysiloxane coatings prepared by the sol−gel process.1 Thewater contact angle of the coating was 37°, confirming that thecoating was hydrophilic.Effect of Plasma Pretreatment of the PC Substrate on
Adhesion to Silica Coating. The adhesion of the dense silicacoating to the PC substrate was evaluated by the ADCB test. Weobserve a significant enhancement of adhesion by shortatmospheric plasma pretreatment, while longer pretreatmentreduces the adhesion gradually. The adhesion energies to the 0,
15, 30, 60, 140, and 270 s atmospheric-plasma-pretreated PCsubstrates are 3.5± 1.2, 4.4± 2.3, 14.5± 2.6, 8.5± 3.3, 6.3± 2.7,and 4.6 ± 1.5 J/m2, respectively (Figure 1). XPS surface analysis
showed that all of the failures are adhesive: the original blank sidehad silicon, oxygen, and carbon elements by XPS, while theoriginal coated side had carbon and oxygen elements. The carbonspecies on the original blank side can be from the coating itself,the developing LMWL, and hydrocarbon and CO2 absorbentsfrom the air. In order to understand the underlying mechanismfor this trend, the chemical and morphological properties of theatmospheric-plasma-treated PC substrate were characterized byXPS and AFM.
XPS Analysis of the Chemical State Evolution of the PCSurface. The chemical states of surface carbon species wereanalyzed to elucidate the chemical interaction between the PCsurface and oxygen atmospheric plasma. The C 1s XPS spectrumof pristine PC is plotted in Figure 2a(i), which can bedeconvoluted to six Gaussian/Lorentzian functions correspond-ing to different carbon bonding states:24−28 (1) 284.5 eV,aromatic C−H; (2) 285.0 eV, aliphatic C−H/C−C; (3) 286.2eV, aromatic C−O; (4) 290.4 eV, O2CO; (5 and 6) 291.3 and292.5 eV, two shakeup satellites for the π→ π* transition of thecarbon atoms in the aromatic ring. [Labeling of the carbon atomsis shown in the inset of Figure 2a(i).] After oxygen atmosphericplasma treatment, two additional peaks appear24−27 [Figure2a(ii)]: (7) 287.5 eV, aliphatic CO and O−C−O; (8) 288.8eV, O−CO.The oxidation of PC follows three steps with increasing plasma
treatment time, as shown by the evolution of the percentage ofthe different carbon components (Table 1 and Figure 2b).Within the first 30 s of treatment, there are sharp drops of thealiphatic C−H/C−C and aromatic C−H species, together withthe diminishing π→ π* shakeup satellites. This suggests that thealkyl groups and aromatic rings are rapidly oxidized or ring-opened by oxygen atmospheric plasma, forming carbonates,esters, O−C−O and carbonyl groups, and oxidized aromaticrings.24−27
At longer exposure from 30 to 60 s, oxidation saturates at thealiphatic C−H/C−C group sites but persists at the aromatic ringsites. More rings are oxidized (decrease of the aromatic C−Hpeak and π→ π* shakeup satellites) or opened (decrease of thesum of the aromatic C−H and C−O peaks). It is also worthnoting that the carbonate and ester species increase faster thanthe carbonyl and O−C−O groups, suggesting that oxidation
Figure 1. Adhesion energy of the dense silica coating on the PC andPMMA substrates with different amounts of atmospheric plasmapretreatment time.
ACS Applied Materials & Interfaces Research Article
proceeds to higher oxidation states at many sites. We believe thatat this stage many PC chains are broken.From 60 to 240 s exposure to atmospheric plasma, the
percentages of all of the components stay almost the same. Theoverall O:C atomic ratio is also stable. This suggests that, at 60 satmospheric plasma exposure, the top 10 nm of the PC surface(the probing depth of XPS is∼10 nm) has reached the maximumoxidation state and many polymer chains have been broken.Further oxidation of these low-molecular-weight oligomersresults in the formation of highly volatile species that wouldeasily leave the surface. In other words, after 60 s of atmosphericplasma treatment, surface etching balances chemical modifica-tion in the 10 nm surface layer (the penetration depth of the X-ray beam). However, the LMWL can grow thicker than 10 nmafter longer plasma treatment, as shown by the AFM resultsbelow. The formation of a LMWL on prolong-treated PC hasbeen reported previously.24
AFM Analysis of the Morphology Evolution of the PCSurface. Similar to the surface chemical state, the surfacemorphology of PC also exhibits changes in three stages. In orderto probe the LMWL that we proposed to form after long plasmatreatments, we compare the AFM images of the plasma-treatedPC before and after rinsing with ethanol (Figure 3). If the LMWL
Figure 2. (a) C 1s XPS spectrum of (i) pristine PC and (ii) 240 satmospheric-plasma-treated PC, deconvoluted to different carbonbonding states. (b) O:C ratio and the area percentage of differentcomponents in the C 1s envelope of PC in relation to oxygenatmospheric plasma treatment time.
Table1.FittingParam
etersfortheC1s
Envelop
eof
PCa
peak
number
12
34
5and6
78
O:C
ratio
bond
arom
aticC−H
aliphatic
C−H/C
−C
arom
aticC−O
O2CO
π→
π*,
arom
atic
oxidized:aliphatic
CO
andO−C−O
oxidized:O
−CO
peak
positio
n(eV)
284.5
285.0
286.2
290.4
291.3,292.5
287.5
288.8
peak
area
percentage
(%)afteratmospheric
plasmatreatm
ent
0s
61.1
18.1
12.2
5.3
3.3
00
0.209
30s
51.6
12.3
20.5
7.9
1.9
1.1
4.7
0.337
60s
48.5
11.3
20.4
8.7
1.7
1.9
7.4
0.404
240s
46.6
11.9
20.4
9.5
1.6
2.6
7.2
0.412
aThe
percentage
ofthedifferentcarbon
speciesandtheO:C
ratio
inrelatio
nto
theatmosphericplasmatreatm
enttim
e.
ACS Applied Materials & Interfaces Research Article
is formed, the surface after rinsing should appear significantlydifferent under AFM.The PC sheet in this study is manufactured by injection
molding from PC resins. The pristine PC surface has hill featuresof ∼40 nm width and ∼1 nm height (Figure 3a), probablyresulting from the inhomogeneity of the melted PC gel at thenanoscale.After 30 s of atmospheric plasma exposure, the height of the
40-nm-wide hills increases from ∼1 to ∼8 nm (Figure 3b),indicating significant surface modification. The flat backgroundhas a higher modification rate than the side walls of the hillsbecause it is perpendicular to the plasma flow and thus receives ahigher plasma afterglow dose per unit area. The surfaceroughness increases from 0.3 to 4.2 nm. The surface morphologyis almost the same before and after ethanol rinse (Figure 3b,c), soit is likely that LMWL has not yet been formed.When the plasma treatment is extended to 60 s, voids begin to
form and larger-sized hills of ∼80 nm width and ∼7 nm height(Figure 3d) replace the 40-nm-wide hills. The LMWL has startedto form, and the very top layer, which contains voids, can berinsed off by ethanol (Figure 3e).There are multiple possible reasons for the emergence of the
larger 80-nm-wide hills. One is that the PC sheet is nothomogeneously cross-linked at the nanoscale. Because the PCsheet is manufactured from the melted PC resins, theintersection region of different resins may be less cross-linkedthan the inner part of the resin because of the different cross-linking conditions for manufacturing the resin and sheet. Thefluctuation of the density and scattering of the reaction center orcatalyst during resin preparation can also lead to inhomogeneity
at the nanoscale. The less cross-linked region of the polymer islikely to form low-molecular-weight oligomers and voids at firstbecause fewer oxidation/disentanglement reactions are required.The 80-nm-wide hills are probably from the more cross-linkedregion. Another possibility is that UV radiation from the plasmacross-links the surface polymer chains and changes themorphology. Cross-linking polypropylene in a surface skin of∼30 nm thickness by oxygen plasma was reported previously.29
Note that melting is not possible at the current conditionsbecause the plasma temperature at the current conditions is 80°C, much lower than the glass transition temperature of PC at148 °C.30 Nevertheless, the 80-nm-wide hills are the main surfacefeatures beyond 60 s of atmospheric plasma treatment, as shownbelow.After 270 s of plasma exposure, a thicker LMWL is formed.
Interestingly, the surface features (Figure 3f) are also hills of∼80nm width. Compared to the 60 s plasma-treated surface (Figure3d), the void-containing layer has been oxidized to volatilespecies. The height of the hills increases from ∼7 to ∼12 nmbecause of a larger plasma dose on the flat background than theside walls of the hills, as discussed above.An ethanol rinse leaves narrower hills of∼40 nmwidth (Figure
3g). This morphology looks like removal of a skin of ∼20 nmthickness from the side walls of the 80-nm-wide hills, leaving the40-nm-wide cores. The hill height increases from ∼12 to ∼20nm, suggesting that an additional 8-nm-thick layer is removedfrom the background compared to the hill side, so the totalthickness of the LMWL on the flat background area is ∼28 nm.The surface skin that is ethanol-soluble provides good
evidence for the formation of thick LMWL by atmospheric
Figure 3.AFM images of PMMA: (a) pristine; after atmospheric plasma treatment of (b) 30, (d) 60, and (f) 270 s; corresponding (c) 30, (e) 60, and (g)270 s plasma-treated and ethanol-rinsed surfaces. Rz is the arithmetic average of the five highest peaks and five lowest valleys in a selected 400 nm × 400nm region.
ACS Applied Materials & Interfaces Research Article
plasma exposure. Forming the LMWL is a combinatorial effect ofthe highly reactive oxygen species and high-energy UV photonsfrom atmospheric plasma. The high-energy UV photons areabsorbed in a polymer surface typically a few tens of nanometersdeep and are sufficiently energetic to break any organic bonds,which can lead to oxidation of the polymer layer,6 especially withhighly reactive oxygen species present.Relating the Surface Chemical State and Morphology
Evolutions to Adhesion Changes of PC. The XPS and AFMresults shed light on the adhesion mechanisms (Figure 4),leading to the trend in Figure 1.At a short exposure time of 30 s, oxidation of the PC substrate
significantly enhances the adhesion to the silica coating. At thisstage, more polar groups are formed through oxidation of thealkyl groups and aromatic rings, which improve the hydrogen-bonding and van der Waals interaction6,7 at the interface.Additional chain ends are also created through chain scission,which increase the interdiffusion of polymer chains and coatingspecies as another adhesion mechanism.6,7 At the same time, thesurface area increases as indicated by the 8 times increase of theheight of the surface hill features, which improve the mechanicalinterlocking of the polymer and coating.6,7
More importantly, covalent bonding6,7 is improved by thenewly formed oxygen functional groups. The oxygen functionalgroups are more likely to form covalent bonds with the coatingthan the more chemically resistant aromatic groups, throughreactions with plasma species, UV photons, and incomingcoating precursors.24,31,32 Last but not least, oxidation has notseverely destroyed the surface polymer network, so the bulksubstrate can achieve strong interaction with the coating throughthe modified interface.At a longer exposure time, as the oxidation of PC proceeds, the
adhesion between PC and the silica coating gradually decreases.In this regime, ring oxidation and opening become predominant,and the surface polymer network starts to degrade to form aLMWL. The LMWL of up to ∼28 nm thickness significantlyreduces the adhesion by hindering the interaction between thebulk PC substrate and silica coating. The formation of a LMWLby long plasma exposure was reported previously7,24 but notcorrelated with quantitative adhesion changes.The reaction kinetics of PC with oxygen atmospheric plasma
gives rise to the initial increasing and then decreasing trend ofadhesion with atmospheric plasma pretreatment time. Shortatmospheric plasma exposure results in the nonaromatic
components being mostly oxidized, forming more polar groupsand chain ends. Because aromatic rings are more chemicallyresistant to reactive oxygen species24,32,33 and UV radiation,34
some of the remaining rings and partially oxidized rings help tomaintain the chain network structure. This site-specific oxidationprovides an exposure time window to obtain a chemicallymodified but not significantly weakened polymer surface toimprove its adhesion to hydrophilic coatings. As oxidationproceeds to more aromatic sites, a LMWL forms and theinteraction between the coating and bulk polymer substrate isweakened, resulting in a decrease in adhesion.
Effect of Plasma Pretreatment of the PMMA Substrateon Adhesion to Silica Coating. The adhesion of the densesilica coatings on PMMA was also evaluated by the ADCB test.As opposed to the adhesion behavior of PC, here the adhesionmonotonically decreases with increasing plasma pretreatment(Figure 1). For the coating deposited on pristine PMMA, thefracture energy is 10.3 ± 1.5 J/m2, comparable to the fractureenergy of the commercial hard coatings on plastics.1 Thecoatings deposited on 5, 17, 34, 170, and 510 s pretreated PMMAexhibit gradually decreasing fracture energies of 9.8 ± 1.4, 9.0 ±1.7, 8.7 ± 1.2, 7.8 ± 2.4, and 3.0 ± 0.8 J/m2, respectively. XPSanalysis of the fractured surfaces shows that all failures occurredadhesively. In order to understand this monotonically decreasingtrend, XPS and AFM were used to study the chemical andmorphological effects of atmospheric plasma treatment onPMMA.
XPS Analysis of the Chemical State Evolution of the PMMASurface. The bonding states of surface carbon atoms wereanalyzed to reveal the chemical effect of atmospheric plasmatreatment on PMMA. The C 1s XPS spectrum of pristine PMMAis plotted in Figure 5a(i), which can be deconvoluted to fourGaussian/Lorentzian functions corresponding to differentcarbon bonding states:35 (1) 284.9 eV, aliphatic C−C/C−H;(2) 285.7 eV, the quaternary carbon atom in the α position to theester group; (3) 286.8 eV, the methyl group carbon single-bonded to an oxygen atom; (4) 288.9 eV, carboxylic O−CO.[Labeling of the carbon atoms is shown in the inset of Figure5a(i).] After oxygen atmospheric plasma treatment, twoadditional peaks appear36,37 (Figure 5a(ii): (5) 287.2 eV, freecarbonyl groups CO; (6) 289.8 eV, carbonate groups O2CO. Maximum oxidation is reached within 10 s of treatment(Table 2 and Figure 5b), compared to 60 s for the PC surface.
Figure 4. Schematic of PC surface modification and associated adhesion mechanisms.
ACS Applied Materials & Interfaces Research Article
The much higher sensitivity of PMMA to oxygen atmosphericplasma is due to the absence of aromatic rings in the backbone.32
The atmospheric plasma modification of PMMA follows onlytwo steps (Table 2). In the first 10 s of treatment, oxidation of theside chain and nonaromatic backbone leads to a dramatic drop of
the aliphatic C−H/C−C species and a significant increase of thecarbon species in higher oxidation states: the ester carbonincreases by 3.1%, and the free carbonyl and carbonate groupsgrow from 0 to 2.2% and 4.1%, respectively. The O:C ratioincreases from 0.385 to 0.640.Beyond the initial 10 s, the area percentages of the different
carbon components become almost constant. The richness ofcarbon in high oxidation states indicates that many polymerchains have been broken to form low-molecular-weightoligomers. At this stage, surface oxidation has reached amaximum and further exposure leads to the formation of highlyvolatile species, resulting in a balance between surface chemicalmodification and etching.
AFMAnalysis of the Surface Morphology Evolution and theRevealed Adhesion Mechanisms of PMMA. The differentadhesion trend for PMMA compared to that for PC can beunderstood by its oxidation kinetics (the previous section) andmorphology evolution, as revealed by AFM (Figure 6).Pristine PMMA has better adhesion to atmospheric-plasma-
deposited hydrophilic silica coatings than the plasma-pretreatedPMMA and pristine PC. The high adhesion is due to plasmaexposure in the initial stage of coating deposition before auniform coating has covered the substrate. Given the coatingdeposition rate of∼110 nm/min, there is direct plasma exposurefor a few seconds. That amount is enough to chemically activatePMMA because 2 s of plasma treatment can already create manynew surface polar groups, which significantly improve the van derWaals, hydrogen, and covalent bonding6,7 with the hydrophilicsilica coating (Figure 7).The difference in surface morphologies of pristine (Figure 6a)
and 5 s plasma-treated (Figure 6b) PMMA also confirms the highsensitivity of PMMA to atmospheric plasma exposure. The largersurface features on 5 s plasma-treated PMMA indicate that manysurface polymer chains have been chemically modified or brokento even out the smaller features on the pristine surface. Thenewly created chain ends can improve adhesion by promotingthe interdiffusion between the polymer chains and coatingspecies (Figure 7). After ethanol rinsing and drying (with N2 gasflow; Figure 6c), the surface morphology changes again,suggesting the flexibility of the surface polymer chains. TheLMWL is not likely to form at this stage because surfaceoxidation has not been completed.When atmospheric plasma treatment extends from 10 to 180 s,
the adhesion only decreases slowly although oxidation of the top
Figure 5. (a) C 1s XPS spectrum of (i) pristine PMMA and (ii) 60 satmospheric-plasma-treated PMMA, deconvoluted to different carbonbonding states. (b) O:C ratio and area percentage of differentcomponents in the C 1s envelope of PMMA in relation to the oxygenatmospheric plasma treatment time.
Table 2. Fitting Parameters for the C 1s Envelope of PMMAa
peak number
1 2 3 4 5 6O:Cratio
bond C−H/C−C quaternary carbon atom inthe α position to the estergroup
C−O O−CO oxidized: CO oxidized: O2CO
peak position (eV) 284.6 285.2 286.3 288.6 287.2 289.8peak area percentage (%) afteratmospheric plasmatreatment
0 s 39.9 20.9 20.3 19.9 0 0 0.3852 s 33.2 20.8 23.0 20.4 1.2 1.4 0.5085 s 31.1 20.2 22.4 21.7 2.4 2.2 0.61310 s 27.9 23.3 21.7 21.0 1.9 4.2 0.64060 s 25.1 24.4 21.2 23.0 2.2 4.1 0.660
aThe percentage of the different carbon species and the O:C ratio in relation to the atmospheric plasma treatment time.
ACS Applied Materials & Interfaces Research Article
10 nm PMMA layer has saturated (the probing depth of XPS is∼10 nm). Interestingly, the surface morphology does not changemuch as well, even after an ethanol rinse (Figure 6d−g),suggesting insignificant residual species accumulation to reduceadhesion. That is because complete oxidation of PMMA takesmany fewer steps than that of PC, and thus the PMMA etch rateis fast enough to prevent significant accumulation of theintermediate oxidation species to form thick LMWL.The slowly decreasing trend can be explained by the
morphology change after 900 s of atmospheric plasma treatment(Figure 6h,i). Over prolonged plasma exposure, the bombard-ment of high-energy plasma species, such as high-energy UVphotons, on the soft and damaged PMMA surface leads to pits oftens of nanometers deep (Figure 5h). A similar pit formationphenomenon was observed during plasma etching of low-kmaterials.38 The deeper etching with pit formation leads toLMWL, which can be removed by an ethanol rinse (Figure 6i).Because pit formation is a slow process, as shown by AFM, it canbe correlated to the slow decrease of adhesion for PMMA withlonger plasma treatment (Figure 7).
Comparison between the PC and PMMA Substratesand with the Literature Report. Upon comparison of thereaction kinetics of PMMA and PC with oxygen atmosphericplasma, it is found that the different sensitivities of the chemicalgroups to reactive oxygen species give rise to different adhesiontrends. For PMMA, the alkyl and ester groups that constitute thepolymer side chain and backbone are both very sensitive toreactive oxygen species, so the side chain and backbone wererapidly oxidized at the same time.13,37 Correspondingly, thesurface is quickly activated, and at almost the same time, LMWLstarts to accumulate. This leads to high adhesion of pristinePMMA to plasma coatings and slowly decreasing adhesion withplasma pretreatment (Figure 7). In contrast, the high resistivityof the aromatic group in the PC backbone decouples surfaceactivation from LMWL formation. By a short plasma exposure,most of the nonaromatic components are oxidized, but thearomatic ring in the backbone helps to maintain the chainnetwork because of its much slower oxidation process. So, asignificant increase of adhesion is observed as a result of surfaceactivation. In the LMWL formation stage that follows, oxidation
Figure 6. AFM images of PMMA: (a) pristine; after atmospheric plasma treatment of (b) 5, (d) 10, (f) 180, and (h) 900 s; corresponding (c) 5, (e) 10,(g) 180, and (i) 900 s plasma-treated and ethanol-rinsed surfaces. Rz is the arithmetic average of the five highest peaks and five lowest valleys in a selected400 nm × 400 nm region.
Figure 7. Schematic of PMMA surface modification and associated adhesion mechanisms.
ACS Applied Materials & Interfaces Research Article
of the aromatic group takes more steps than that of the alkyl andester groups, so more oxidation intermediates are accumulated toform thick LMWL, resulting in a sharp drop in adhesion (Figure4).This comparison suggests that constructing a less oxygen-
sensitive backbone, such as incorporating aromatic rings in thebackbone, is important for using atmospheric plasma to enhancepolymers’ adhesion to hydrophilic coatings. In this way,oxidation of the polymer side chain and backbone can bedecoupled to different time scales, allowing for a time window toform a substrate surface rich in polar groups and chain ends witha maintained chain network.It is also worth noting that using higher plasma powers of up to
80 W to pretreat the PC and PMMA substrates resulted in thesame trends but at shorter times. However, when the plasmapower was too high (≥100 W for our system), the plasmatemperature increased to above the glass transition temperatureof the polymer substrate.There are some previous studies about the effect of oxygen
plasma treatment of aromatic polymers and PMMA12,13,39 ontheir adhesion to epoxy and resins. In one study usingatmospheric oxygen plasma,11 the adhesion of aromatic polymersto epoxy increased after plasma treatment. Chain scission11,12,40
and LMWL formation11,24 were observed, but no adhesion dropafter longer treatment was reported. In another study,7 whichcompared the treatments of low- and atmospheric-pressureplasmas, the authors reported that the low-pressure plasmatreatment increased the adhesion with extended treatment timebut prolonged atmospheric-pressure plasma treatment decreasedthe adhesion to epoxy. We believe that these differentobservations result from the nature of the epoxy and thethickness of the LMWL. If the epoxy can penetrate the LMWL tointeract with the bulk substrate, the adhesion increases withincreasing treatment time. Conversely, if the epoxy cannotpenetrate LMWL to bond well with the bulk substrate, adhesiondecreases. In our case, we deposited plasma coatings onpolymeric substrates. It appears that the gas-phase-depositedcoating is sensitive to the presence of LMWL, which significantlyhinders interaction between the coating and bulk substrate andreduces adhesion.
■ CONCLUSIONOxygen atmospheric plasma pretreatment was explored toimprove the adhesion of PC and PMMA to plasma-depositedhydrophilic silica coatings. The treatment time and polymerchemical structure are found to be important factors. For PC, ashort treatment of 30 s increased the adhesion energy by a factorof 4, while longer treatment decreased the adhesion gradually. Incontrast, a monotonically decreasing trend of adhesion withlonger plasma treatment was observed for PMMA, with thestarting adhesion energy much higher than that in the PC case.The chemical state and morphology of the atmospheric-
plasma-treated surfaces were characterized to understand thesetrends. At short exposure times, the formation of surface polargroups, additional chain ends, and increased surface roughnessare found to improve adhesion through covalent-bonding,hydrogen-bonding, and van der Waals interactions, chain/coating species interdiffusion, and mechanical interlocking forboth PC and PMMA.If the polymer is highly sensitive to oxidation, which is the case
for PMMA, plasma exposure during the initial stage of plasmacoating deposition can be sufficient to enhance adhesion. If thepolymer is more resistant to reactive oxygen species, for instance,
because of the presence of aromatic rings in the PC backbone, amoderate atmospheric plasma pretreatment can decouplesurface activation from LMWL formation and significantlyenhance adhesion. Nevertheless, if the treatment time is too long,overoxidation of both the PC and PMMA surfaces can happen,which leads to LMWL formation and reduced adhesion. Theseresults point out the importance of the plasma pretreatment timeand polymer chemical structure for adhesion enhancement.
■ ACKNOWLEDGMENTSThe work was supported, in part, by the Director, Office ofEnergy Research, Office of Basic Energy Sciences, MaterialsSciences Division of the U.S. Department of Energy, underContract DE-FG02-07ER46391, and by the Boeing Co.
■ REFERENCES(1) Kamer, A.; Larson-Smith, K.; Pingree, L. S. C.; Dauskardt, R. H.Thin Solid Films 2011, 519, 1907−1913.(2) Erlat, A. G.; Spontak, R. J.; Clarke, R. P.; Robinson, T. C.; Haaland,P. D.; Tropsha, Y.; Harvey, N. G.; Vogler, E. A. J. Phys. Chem. B 1999,103, 6047−6055.(3) Zhu, P. X.; Teranishi, M.; Xiang, J. H.; Masuda, Y.; Seo, W. S.;Koumoto, K. Thin Solid Films 2005, 473, 351−356.(4) Brown, M. D.; Suteewong, T.; Kumar, R. S. S.; D’Innocenzo, V.;Petrozza, A.; Lee, M. M.; Wiesner, U.; Snaith, H. J. Nano Lett. 2010, 11,438−445.(5) Lionti, K.; Toury, B.; Boissier̀e, C.; Benayoun, S.; Miele, P. J. Sol−Gel Sci. Technol. 2013, 65, 52−60.(6) Liston, E. M.; Martinu, L.; Wertheimer, M. R. J. Adhes. Sci. Technol.1993, 7, 1091−1127.(7) Shenton, M. J.; Lovell-Hoare, M. C.; Stevens, G. C. J. Phys. D: Appl.Phys. 2001, 34, 2754−2760.(8) Kruse, A.; Kruger, G.; Baalmann, A.; Hennemann, O. D. J. Adhes.Sci. Technol. 1995, 9, 1611−1621.(9) Goldman, M.; Goldman, A.; Sigmond, R. S. Pure Appl. Chem. 1985,57, 1353−1362.(10) Klemberg-Sapieha, J. E.; Martinu, L.; Sapieha, S.; Wertheimer, M.R. In The Interfacial Interactions in Polymeric Composites; NATO-ASI,Series E: Applied Sciences; Akovali, G., Ed.; Kluwer: Dordrecht, TheNetherlands, 1993; Vol. 230, pp 201−222.(11) Gonzalez, E.; Hicks, R. F. Langmuir 2009, 26, 3710−3719.(12) Gonzalez, E.; Barankin, M. D.; Guschl, P. C.; Hicks, R. F.Langmuir 2008, 24, 12636−12643.(13) Gonzalez, E.; Barankin, M. D.; Guschl, P. C.; Hicks, R. F. PlasmaProcesses Polym. 2010, 7, 482−493.(14) Gonzalez, E.; Barankin, M. D.; Guschl, P. C.; Hicks, R. F. IEEETrans. Plasma Sci. 2009, 37, 823−831.(15) Pappas, D. J. Vac. Sci. Technol., A 2011, 29, 020801.(16) Cui, L.; Ranade, A. N.; Matos, M. A.; Pingree, L. S.; Frot, T. J.;Dubois, G.; Dauskardt, R. H. ACS Appl. Mater. Interfaces 2012, 4, 6587−6598.(17) Jeong, J. Y.; Park, J.; Henins, I.; Babayan, S. E.; Tu, V. J.; Selwyn, G.S.; Ding, G.; Hicks, R. F. J. Phys. Chem. A 2000, 104, 8027−8032.(18) Foest, R.; Kindel, E.; Lange, H.; Ohl, A.; Stieber, M.; Weltmann,K. D. Contrib. Plasma Phys. 2007, 47, 119−128.(19) Li, G.; Li, H.-P.; Wang, L.-Y.; Wang, S.; Zhao, H.-X.; Sun, W.-T.;Xing, X.-H.; Bao, C.-Y. Appl. Phys. Lett. 2008, 92, 221504.(20) Kanninen, M. F. Int. J. Fract. 1973, 9, 83−92.
ACS Applied Materials & Interfaces Research Article
(21) Suo, Z. G.; Hutchinson, J. W.Mater. Sci. Eng., A 1989, 107, 135−143.(22) Anderson, T. L. Fracture Mechanics: Fundamentals andApplications; CRC Press: Boca Raton, FL, 2005.(23) Nowling, G. R.; Yajima, M.; Babayan, S. E.; Moravej, M.; Yang, X.;Hoffman, W.; Hicks, R. F. Plasma Sources Sci. Technol. 2005, 14, 477−484.(24) Muir, B. W.; McArthur, S. L.; Thissen, H.; Simon, G. P.; Griesser,H. J.; Castner, D. G. Surf. Interface Anal. 2006, 38, 1186−1197.(25) Andrady, A. L.; Searle, N. D.; Crewdson, L. F. E. Polym. Degrad.Stab. 1992, 35, 235−247.(26) Andrady, A. L.; Fueki, K.; Torikai, A. J. Appl. Polym. Sci. 1991, 42,2105−2107.(27) Clark, D. T.; Dilks, A. J. Polym. Sci., Polym. Chem. Ed. 1977, 15,2321−2345.(28) Lannon, J. M.; Meng, Q. Surf. Sci. Spectra 1999, 6, 75−78.(29) Schonhorn, H.; Ryan, F. W.; Hansen, R. H. J. Adhes. 1970, 2, 93−99.(30)Markolon polycarbonate data sheet, BayerMaterial Science, 2013.(31) Vollhardt, K. P. C.; Schore, N. E.Organic Chemistry: Structure andFunction, 6th ed.; W. H. Freeman: Cranbury, NJ, 2011.(32) Groning, P.; Collaud, M.; Dietler, G.; Schlapbach, L. J. Appl. Phys.1994, 76, 887−892.(33) Koprinarov, I.; Lippitz, A.; Friedrich, J. F.; Unger, W. E. S.; Woll,C. Polymer 1998, 39, 3001−3009.(34) Chapiro, A. Radiation chemistry of polymeric systems; Interscience:New York, 1962; p 446.(35) Girardeaux, C.; Pireaux, J.-J. Surf. Sci. Spectra 1996, 4, 134−137.(36) Ben Amor, S.; Baud, G.; Jacquet, M.; Nanse, G.; Fioux, P.; Nardin,M. Appl. Surf. Sci. 2000, 153, 172−183.(37) Chai, J. N.; Lu, F. Z.; Li, B. M.; Kwok, D. Y. Langmuir 2004, 20,10919−10927.(38) Darnon, M.; Casiez, N.; Chevolleau, T.; Dubois, G.; Volksen, W.;Frot, T. J.; Hurand, R.; David, T. L.; Posseme, N.; Rochat, N.; Licitra, C.J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.Process., Meas.,Phenom. 2013, 31, 011207.(39) Gonzalez, E.; Barankin, M. D.; Guschl, P. C.; Hicks, R. F. IEEETrans. Plasma Sci. 2009, 37, 823−831.(40) Hofrichter, A.; Bulkin, P.; Drevillon, B. J. Vac. Sci. Technol., A2002, 20, 245−250.
ACS Applied Materials & Interfaces Research Article
